On the Use of Recursive Partitioning in Causal Inference: A Proposal

A tree-based method for identification of a balanced group of observa- tions in casual inference studies is presented. The method derives from an algorithm which uses a multidimensional balance measure criterion to recursively split the dataset based on the values of the covariates. Observations are finally partitioned in subsets characterized by different degrees of homogeneity. An ad-hoc resampling scheme is used to select the units for which causal inference can be carried out.